Slashdot Mirror
Electronic Circuit Mimics Brain Activity
A lot of people wrote in with the news blurb from Yahoo! regarding the announcement of a ciruit that supposedly acts in a manner resembling human brain activity. Details in the blurb are pretty sketchy though - post links below if ya got 'em. One of the interesting points that they say though is that the brain does both digital and analog - but that's pretty much all they say about it.
Check out the original article, in the letters to nature section. Rodney Douglas' web adress www.ini.unizh.ch/~rjd/
Enjoy
Jan-Jan
Everything is "analog". The binary encoding ("digital") is just bolted over the analog thing, think for example a voltage signal with, say, NRZL encoding. It's digital over analog.
BUT when you go deep enough, you get discreteness, quantum states and the digital 1/0 is/not_is yin/yang world again appears! A most beautiful cycle!
Ever seen a nice image of a neuron? It has an input end, and an output end. A neuron may also have multiple output ends, and multiple output ends may come together to one neuron's input end.
Neurons don't only "decide" to fire or not to fire... They can fire at a multitude of magnitudes. All neurons have a trigger level. An incoming pulse needs to be of that magnitude or higher for the neuron to react. The neuron can then "decides" to fire, but neurons can also increase or decrease the magnitude of the pulse they send off. That's the way a neural net works. Various pulses of various magnitudes travel through a neural net, and according to the magnitude of a pulse, the receiving neuron "decides" what to do with it.
You could indeed consider the "decision" to fire or not to fire a binary decision. Compare it to a joystick. It's not a binary joystick that either goes full or not at all, but it's one of those analog joysticks that can go everywhere inbetween aswell. The neuron can have dead zones on both ends, can alter sensitivity, can switch axis and could even reverse axis. On the whole, neurons adapt and learn. All these "neuron settings" cause various pulses to travel different paths through the same network, to split, to die off, and do all sorts of other interesting stuff.
)O(
the Gods have a sense of humour,
Never underestimate the power of stupidity
To err is human, to moo bovine
IANAB (I am not a biologist), but I do seem to recall that neurons are basically onion-shaped, so to say, with a thick roundish end where the cell core and other such useful things are, and a long "tentacle" that stretches out. I also seem to recall that that "tentacle" splits into various others, to (almost) touch various other neurons at the receptors on the thick end. Then again, the last time I had biology is four years ago, and then most of that was all about (populational) genetics, (natural) selection and the calculations behind those... :)
)O(
the Gods have a sense of humour,
Never underestimate the power of stupidity
To err is human, to moo bovine
OK, I'm an ex-neuromorphic VLSI researcher, so my impressions may be colored, but let's see...for the last ten years, we've been following Carver Mead's lead that we really need to look at Analog VLSI for simulating cortex and doing cool AI work. It's ulta-low-power, distributed massively parallel computation, defect tolerant, etc. And what has been the result?
Millions of dollars of money going into making bad retinal focal-plane arrays whose output makes QuickCams look good, analog cochlea models that underperform real time digital models, and a handful of other do-nothing circuitry like the one described in this article.
Meanwhile good old digital VLSI has gone from 100 MHz to > 1 GHz, we have actual speech recognition systems running on PCs, and a new range of low-power digital multi-purpose digital CPUs for portable devices.
There has never been a real product developed using neuromorphic VLSI, and the few implementations can now be replaced with faster digital computers.
The best part of neuromorphic VLSI was the electical engineers teaching all the neuroscientists about how electical circuits work, wavelet transforms, etc., to a bunch of people who like to think of the brain more in terms of a Rube Goldberg device where one neuron taps the next neruon instead of a complex chaotic set of electrical network equations.
thinks more like humans? That is, a computer that repeats unreliable gossip, makes wild guesses when it doesn't know the answer, tries to cover up it's mistakes and blames the computer in the next cube, calls in sick when there's something better to do, procrastinates, has hidden agendas, is married to a plain wintel box but secretely is in love with the well endowed cluster in R&D, worries about it's retirement plan, thinks it deserves a raise and the corner office but is discriminated against by office politics...
try { do() || do_not(); } catch (JediException err) { yoda(err); }
There is a old branch of EE called neural networks.
This is merely a variant of of it.
One of the more interesting neural networks is
CalTech's Carver Mead's artificial retina.
It has excitory and inhibitory connections.
His company put hundreds of thousands of them
on a chip to make a self-regulating and processing
digital camera. Got a White House medal for this
and other stuff.
There are a trillion neurons in a mammalian brain
with a thousand connections each. It will take a while to emulate this.
Good mingling of Consciousness and QM here:
here
I can't speak for his physics (he gets into manifolds and such) but the ideas are right on.
Just another perl hacker in Bangkok
...not what we're looking for. We aren't looking for a statement that a human can't assert. We're looking for a statement that is true (a condition yours meets) but that a human can't know is true.
This maps to, for instance, number theory as follows: There are theorems that are true (that is, we know that they state things as they really are) but that can't be derived (i.e. proven) with the axioms and rules of number theory. This is true no matter what axioms you use: even if you add those theorems new ones are always "out there".
It's easy (well, relatively) to find these statements in other systems, but it may be logically impossible for US to find them in the human brain. Maybe aliens (or AI) will have to prove that Godel applies to us (and us to them).
Hey! That gives me an idea: Maybe "Godel's theorem applies to humans" is the human godel statement.
--
Linux MAPI Server!
http://www.openone.com/software/MailOne/
(Exchange Migration HOWTO coming soon)
"It consists of artificial neurons that communicate with each other via synapses, or junctions where they connect, in a system that could lead to the development of computers that could perform perceptual tasks such as sight recognition."
That's absolutely incredible. Think of the power we now wield: We can transport scientists through time from the 1950's.
We are able to "carbon date" these scientists by means of the research they are doing. For instance, had they been attempting to determine the speed of the planet Earth through the "luminiferous ether" we would have known they came from before 1903. Had they stared in wonder at our televisions, we would have known they pre-dated the 1950's. However, their work on the then cutting-edge, now old-hat neural networks (implemented in hardware, no less) places them firmly in the 1950's.
--
Linux MAPI Server!
http://www.openone.com/software/MailOne/
(Exchange Migration HOWTO coming soon)
I think you might be confused over what digital means, its not a specification for an electric wave.
"A digital system is a set of positive and reliable techniques (methods, devices) for producing and reidentifying tokens, or configurations of tokens, from some prespecified set of types" (from "Artificial Intelligence", Haugeland)
A positive technique is one that can succeed absolutely. You can write something in sloppy handwriting, but i can still (potentially) succeed absolutely in reading what it is that you have written. The potentially is important, because a digital system is not required to be reliable, just that it is possible to work absolutely correct. The mechanics of how a brain can read arbitrary handwriting (or any number of things we do everyday) is undoubtedly analog, but some systems within the brain are digital. Writing a post for slashdot is analog, i cannot "absolutely succeed" in putting my words together in an intelligible way that says everything a post should. But the act of posting a post is digital, i can succeed absolutely by pressing this little submit button down here, which i will do....now
Chaos theory doesn't demonstrate much of anything about quantum mechanics. The mathematics of chaos work perfectly well with our friends the integers and the reals; no exotic particles required. Sorry.
Heisenberg's work isn't part of chaos theory, bucko. Check out the excellent Chaos Page at the University of Maryland.
:-)
Chaos theory deals with "simple nonlinear deterministic systems" that "behave in an apparently unpredictable and chaotic manner".
Sorry, but quantum mechanics doesn't count.
Although now that I think about it, many Anonymous Cowards are simple detministic systems that behave in a chaotic manner. Maybe you should go in as a lab animal.
"The brain processes both analog and digital signals."
"... the brain makes an either-or decision about whether or not it is a car"
At first, I thought "How are these ideas so different from eg. a scanner reading some text (analog) which is then OCR'd by software which decides whether it sees a letter A for example (either-or digital decision)
But of course the important difference here is that the brain processes analog and digital together. All existing electronics (before this new research) processes anolog and digital completely seperately with just an interface between the two. The scanner is the interface in my example.
Can anyone think of better existing examples with both analog and digital components but where the anaog-digital connection is more intimate?
-- MartinG To mail me: echo kewyjlcxyzvjfxbqwh | tr bcefhjklqvwxyz
I haven't read either book, but...
I could understand a point of view that said "We don't know if quantum effects manifest themselves on a macroscopic level in the brain".
I'm having difficulty with the idea that quantum effects might be the only way that some brain processes could possibly function.
I have heard some people refer to quantum brain effects as some kind of new age "soul"... As a devout atheist that rubs me up the wrong way.
*sigh*
I believe our friend was saying that due to sensitive dependence on initial conditions (or state divergence) exhibited by many nonlinear systems, the effects of quantum events aren't simply averaged out in the macroscale and swallowed by the law of large numbers.
The butterfly effect for nonlinear systems like weather, or parts of our own brains, should apply right down to the quantum scale.
Computing consiousness may require quantum algorithms as Penrose indicates.
My quantum tentacle presses the Submit button...
"It is almost a car," or, "It might be a car," implying that there is a degree to which something might be a car."
This is nonsense IMO. "This is almost a car" is just as much a digital statement as "this is a car". The brain can be 100% positive about its own uncertainty.
As an aside, the spike trains generated by neurons are purely digital. Only the strength of synapses are analog and this strength can be easily simulated digitally by a sufficiently large integer. All this "analog is superior" crap is just that, crap.
Louis Savain
It isn't all that new. It's an evolving field that Carver Mead basically started back in the 80s that's slowly been building steam. The problem is that while the cost for neurons in biology is very low, that's not the case in silicon, so building large scale systems is problematic. Further, the compuations tend to be analog in nature, so you have to take care how you design the cirucits. Still, the area is a neat thing to watch evolve.
The linux types around here should like the approach they take: free tools they wrote (and distribute) that run under linux. Try http://www.pcmp.caltech.edu for the details on their work.
Will they suffer from alzhiemers?
If I make my computer a duel boot will it be suffering from multiple personality disorder?
Wheeeee
When did your friend visit? I was at Penn in December of 1998 interviewing for a graduate program, and the system I saw didn't require a whole bunch of switch-flipping; it was just a bunch of standard DIPs interconnected on a breadboard and hooked to an apparently standard video camera. The result did appear to model what I know of the initial visual processing system.
Of course, any smoothly functioning technology is indistinguishable from a rigged demo, but I doubt they'd go to the trouble of setting up a flashy-but-fake demo to impress a prospective grad student. I'm just not important enough to lie to.
Specific papers probably aren't the best way to get an good overview. My personal textbooks are _Cognitive Psychology_ by Medin and Ross and _Neuroscience_ by Bear, Connors, and Paradiso. However, these are not ones I specifically chose; they're the textbooks chosen by my professors. You may have better luck just searching for popular stuff at Amazon.
Note that the different encodings thing is not proven; it merely makes sense, since cells in different brain areas often have significant phenotypic variation.
To your question about forgetting... current theory says it's more of your second hypothesis. In this regard, the brain appears to act like a neural net --- if you train it to do something, but then start training a different task and never provide the stimuli for the first task as reinforcement, it'll drift away from its intial conditioning.
Current theory also views memory sort of like a hash table: the entire state of the system is the input, and something comes out based on associations. This leads to an effect called "state-dependent learning": if you learn all your facts sitting in one desk of a classroom, you'll do better on the test if you take it at that same desk. As you age, your sensory inputs change, which means it's harder to construct a world-state capable of accessing a given piece of information. This has been suggested as the reason why most people can't remember their childhoods well --- our growth has changed the way the world looks to the point that we just can't construct an activating signal for anything but the strongest memories.
Connections are definitely not permanent. There is basically nothing permanent about the brain beyond the gross organization of areas and layers. The absent-minded professor effect, IMHO, is more a matter of changes in significance. You remember the things you think are important. As you get to concentrating more and more on proving P=NP, little details like when you last ate just aren't relevant enough to be encoded. (For most people, memory seems to have some sort of finite bandwidth, such that only the most significant aspects of the current state are likely to be encoded.)
I believe it's nothing as severe as multipolar to bipolar/unipolar, but yes, the basic mechanism is believed to be a rearrangement of cellular projections. Neurons can also modulate synaptic weights by messing with the balance of channel proteins, vesicle docking proteins, and other key pathway components. To the best of my knowledge, neurons tend not to apoptose once the brain reaches maturity, although there are massive die-offs early in life as the pathways get themselves sorted out and unnecessary cells get pruned. There are some exceptions (I know that some cells of the olfactory pathway are regularly dying and regenerating, and I believe taste cells do the same thing), but for the most part neurons are pretty long-lived beasts. Since they're nonmitotic and the stem cells tend not to produce more, it's to the organism's advantage to conserve neurons.
At the University of Delaware we've had a complete system setup with emulated neural circuitry for some time now. Each circuit is a hybrid analog/digital artificial neuron called a "neuromorph".
The spikes are recorded by a seperate board and routed through hardware buffers to "synapses" on the next circuit, thus emulating the "leaky-integrate and fire" mechanisms of neurons.
For more information e-mail me at the above address (yes it's real) and I can point you to research articles and information that has been published from our Neuromorphic Systems Laboratory at Udel.
Even so, this is not a new thing, the theory behind artificial neural networks dates back some 40+ years, and there have been many attempts at Universities to implement the most realistic and interesting mimicries of human behavior.
It's all analog. From what I've been told by someone who's been there, they have to flip all kinds of switches to make the networks and they act rather stagnantly because it takes so much work to change it.
There's alot of debate over which system provides the most realism vs. the most flexability. I think the answer lies in several Universities' approaches (including Udel and apparently MIT's new setup) as an analog-digital hybrid.
Neuromorphic Engineering is a real term to describe this kind of research. The problem is, it isn't new. (See some of my other comments...)
What I learned about neurons is that they are all or nothing. If the stimulation is not enough, there is no firing, else it fires completely. In that sense it is digital. I think the analog part comes in on the rate of firing. There is a certain limit on how fast a neuron can fire (ie after the action potential was raised to its limit and is now lowering, it cannot fire again) but other then that information is conveyed in the analog frequency of firing.
Just because a baby's response _acts_ like objects either exists or don't, doesn't necessarily means that the thought-processes of a baby (both conscious and unconscious) is limited by this. We don't really know all the thoughts of a baby -- not even our own thoughts. Just think of how much thoughts is hidden down in our brains, for example hidden trauma. The way we deal with these is to dream about them.
I can think of another answer to this: That the extreme short attention-span of babies make it more efficient for them to learn basic things faster. As we grow up however, we need attention-span to grow and become more abstract-thinking, to learn and reflect on more complex things. This is less efficient on more basic problems however (more overhead).
I might be wrong, but so might 1.000 scientists.
- Steeltoe
http://www.debunkingskeptics.com/
Well it sounds to me that that "final estimate" step if it exists at all is just a way to destroy the whole result! If someone is able to know the days of any date instantly (I have talked to a person like that), why would they want to degrade this into an "estimate"?
....." to the result set. Or you might use this to every operation involved to get an even more fuzzy estimate.. :-) A machine can have "feelings" too, in the same way. However, I believe anything we do with a machine is just a poor emulation on what's really going on in our brains. But it's still an emulation, and I believe it is possible to develop it so that we won't be able to distinguish it from a live person. Not in the near future though. (What are we really trying to create with neural nets? Copies of ourselves? Why not just procreate?)
Yes, machines can estimate: For any result X just do an Y = X + random(d)-d/2, and add "I don't really know for sure, but I think it's
Now, it seems we humans are dependent on being able to estimate things. It's a role of being flexible and adaptable. We have logic, but it's very fuzzy. This is an disadvantage when dealing with "digital" datasets, but not so when we're living our daily lives.
If we can trigger our "rainman" capabilities inside our brains and harmonize this with what we already got, will we ever need a computer again?
- Steeltoe
http://www.debunkingskeptics.com/
I saw something very interesting related to this at the University of Technology Sydney (UTS or uterus as it is often called;)) open day. They were working on a project in which disabled people would be able to use the part of the brain related to that disabilty for some use. For example if you were blind then they would work on detecting the thought processes for sight so that one thought may trigger a circuit to turn on a heater and another thought may for example turn on an oven. Pretty interesting and a lot of potential if you ask me.
This should not be (4: Funny). It violates the standard form of the "Can you imagine..." post, giving away the punchline in the subject line. Please moderate it down.
Karma: Good (despite my invention of the Karma: sig)
Just use open source, they would play QUITE nicely together
If you don't understand, go look up Salon.coms article on sex and hackerdom from a couple weeks ago...
I'll just wait for my trinary!(yes/no/maybe)
I did notice that these guys have been tinkering round with neural stuff for a while. I found this article which is interesting and along a similar vein and has a pretty picture in it, or here which is the press release without pretty pictures.
I off to book a holiday at Westworld now.
"Common sense is nothing more than a deposit of prejudices laid down in the mind before you reach 18" Einstein
The way the article used the car example (saying our brain says it either is or it isn't a car) is the same as saying that my odds of winning the state lottery are half because either I win or I don't.
So, what they've done is taken standard Nueral Net technology that has previously been implemented in software, and engineered customer hardware to do it instead. Cool. I guess that means Nueral Nets can work much faster than before, which is nice.
But (And there has to be a but), the way i understand it is that with SW Nueral Nets, synapsis can grow and die off as the pathways are enforced in the net. If the Neurons and Synapsis are hardwired, doesn't this limit the ability of the Net to grow? What happens when all of the available synapsis are currently in use? Just swapping in a new chip with more synapsis on it isn't an answer, the new chip would have to re-train to do the same job that the older one did. So, are hardware Neural Nets a real advantage over Software Nets?
Syllable : It's an Operating System
It seems to me that scientists are working this thing in the wrong direction. I mean, why are they trying to figure out the (to our knowledge) most complex brain in the world? Isn't that like starting your studies as a computer engineer with trying to reverse engineer an Athlon processor or the equivalent?
Wouldn't it be a better idea to first try to fully understand and map out a very small brain like, let's say, a bee or something similar? Their brains sure perform lot's of functions (Like before mentioned image recognition), but there is much less brain cells and synapses and stuff to examine. Then they could work their way up with more and more complex brains.
Just like the ppl who mapped the human genes have done, they started with simple flies...
--
"I'm surfin the dead zone
--
"I'm surfin the dead zone
In the twilight, unknown"
--Fesh
--Fesh
Kill -9 'em all, let root@localhost sort 'em out.
If the goal is to create intelligent perceptual machines, why go backwards and design like a human brain? Why not go forwards and design according to the desired function at hand. The human brain is not the most efficient computing substrate for many tasks.
-- Matthew - matthew.gream@pobox.com, http://matthewgream.net
(stanislav) Grof`s book `beyond the brain` has stuff about how the brain might work, but i dont think anyone knows yet...
Tell that to Aphex Twin.
If the two are different, then one can't behave exactly the same as the other -- that's what different means.
---
script-fu: hash bang slash bin bash
[ approaching AI ]
Analogue circuits do stuff that digital ones don't. Digital data is more preservable, yes, but it's less accurate, it fits things into integrals... into little tiny boxes -- analogue doesn't do this at all... it's a true medium, where the signal is properly 'carried.' The advantage of digital is the the medium won't distort the message.
To get digital accuracy such that you can't tell the difference is like trying to represent Pi as a fraction. Honestly, to be fully accurate you'd need an infinite number of signal samples.
---
script-fu: hash bang slash bin bash
[ approaching AI ]
I already mentioned this. The real point is analogue signals and digital signals are different, and you can't use a digital signal to replace an analogue one.
---
script-fu: hash bang slash bin bash
[ approaching AI ]
What if the brain sends signals in parallel, and not in serial, it could seem that they are then sent as analog? I am pretty sure there are many things we don't know about the brain, this could be one of them....
-
As cunning as a fox, which has just been appointed professor of cunning at Oxford University. http://www.kinlan.co
grammar nazi sez: I still believe that humans make analog thoughts, even if our brain is just one big circuit. Can I use our brain tonight?
For all those scientists out there who love wading through scientific publications, my old employer, the Mental Health Research Institute has a department dedicated to Brain Dynamics. Some of their published papers are available HERE
Their main goal was to simulate brain processing in software rather than hardware.
Good boy, good boy, here is your candy. Thank you
for alerting us of such a terrible terrible crime.
Luckily we have you, otherwise we might stress our
brains too much. You really are mans best friend.
Uum, no -- what the hell are you talking about?
Did you even read the article? They are talking about the internet as a collectively intelligent network of computers -- not just intelligent computers.
Even so, how does that compare to Terminator 2??? A military computer controlling weapons-of-mass-destruction achieves sentience and decides human beings are its primary threat? Do you seriously think that military computers close to important military weapons, etc. are hooked directly to the internet to even make this scenario even remotely possible?
Even if that were true I'd be more worried about hacker/cracker terrorists than a computer-based threat! Or how about an asteroid? Or how about lightning? Man, it's fearful people like you who make the adoption of technology so ridiculously difficult. Sure be skeptical, but don't TRY to come up with these ridiculous scenarios, use some logic.
So what? A bowl of lime jello hooked up to an EKG gives the same readings as a human brain. Weird, but true, I think.
-JT
Shameless TUGHouse Plug
It isn't that simple. I am unfortunately not a neurologist, but I did study brain mechanics for two semesters of Psychology, so please bear with me.
If I remember correctly, the way it works is the neuron gets signals from many other sources along its dendrites, which either have an inhibitory or excitatory response, which also decays along the way (the "analog decision"). The decision to fire or not at any given time is digital to some degree, but there are factors that affect that as well such as the refractory period (an axon may only depolarize and fire once every so often).
However, once the electrical impulse reaches the end of the axon, it does not leap across to the next dendrite. It releases chemical agents which float across to the dendrite and interact with it, which are then recovered by the axon. Those chemical agents vary from axon to axon.
In addition, the recovery (reuptake) of those chemicals is crucial. What if a drug affecting the brain prevents the reuptake of those chemicals? Then they start floating around and reacting with any dendrite they happen to run into, and they get the same effect as if an axon had been fired. In fact, this is part of what happens when you become inebriated... Dopamine reuptake is blocked, causing it to float about in your brain and get overused, making you get a buzz.
Unfortunately, there is a limit to what I can remember. There are seven ways that a drug can interact with a neuron to create an inhibitory or excitatory response, however. So while it may seem digital, there is a lot more to the human brain than "fire/not fire".
Dave
I don't agree 100% that the brain makes digital decisions. The article says that we make an either/or decision regarding whether something is there or not. It is a car or it isn't a car. That's rather black and white. If a picture is blurry or if the object is partially hidden, then we could say, "It is almost a car," or, "It might be a car," implying that there is a degree to which something might be a car.
This is another reason why the brain is so difficult to emulate... When a human makes a decision like that, he or she uses a combination of bottom up and top down processing. The bottom up processing sees shapes and lines (such as simple things like vertical or horizontal lines, or maybe even a more complicated shape such as a triangle) and builds the image from that. However, at some point the top down processing steps in and says "Hey, that kind of looks like a car. So it must be a car." The entire process is not built up from scratch every time you look at an object.
In conclusion, you are correct... Not only is it both digital and analog to some degree but there seems to be a lookup table of some sort created. Even more mysteries to unravel.
Dave
MIT, bringing you the Butlerian Jihad one discovery at a time.
Dave
Here I found links to the home pages of Sebastian Seung and Rahul Sarpeshkar, two guys mentioned in the Yahoo article. A quick look doesn't reveal much specific to this story but, not surprisingly, all there research is in this area.
Hrrrm .... it's an electronic circuit .... it's a Universal Turing Machine .... it isn't like a human brain .... next dull Reuters story, please, nurse, I can still feel my legs ....
-- the most controversial site on the Web
This article from this week's New Scientist covers similar ground - inteligent computers. The concept here is that the internet will become inteligent in it's form as a distributed network.
Does that sound like the movie Terminator 2 to anyone else?
(Spudley Strikes Again!)
I wonder how long until we have the first bipolar computer.
Mimics human brain activity eh!
This must be a major step towards developing artificial idiocy
134340: I am not a number. I am a free planet!
As somebody has already pointed out, nueral nets are old hat in AI. And not all of them are computational models of neural nets, some are actually embedded. I've used a simple neural net in a stiqueto-based robot to control it's leg movement. The only way a hardware based model would be new is if it actually emulated the brains electro-chemical reactions instead of using digital logic. The problem with hardware implementations (as opposed to computer modeling is:
Every neuron has axon's connected to other neurons. This is called Fan-out. In the human mind, average fan-out is somewhere along the lines of 1000, as obossed to an artifical net, which has a avereage fan-out of 10. The problem with implementing this in hardware, even with a fanout as little as 10, is attenuation. Every time the neuron fires, the signal looses strenght. Because real nuerons are electro-chemical, and are self-powered. they don't have this problem. In a small net (like in my stiqueto-bot), i just used a couple a 'strategically' placed op-amps (my strategy involved poking it with a meter =). But in a net with a fan-out of 1000. This wouldn't be practical.
As somebody has already pointed out, nueral nets are old hat in AI. And not all of them are computational models of neural nets, some are actually embedded. I've used a simple neural net in a stiqueto-based robot to control it's leg movement. The only way a hardware based model would be new is if it actually emulated the brains electro-chemical reactions instead of using digital logic. The problem with hardware implementations (as opposed to computer modeling is: Every neuron has axon's connected to other neurons. This is called Fan-out. In the human mind, average fan-out is somewhere along the lines of 1000, as obossed to an artifical net, which has a avereage fan-out of 10. The problem with implementing this in hardware, even with a fanout as little as 10, is attenuation. Every time the neuron fires, the signal looses strenght. Because real nuerons are electro-chemical, and are self-powered. they don't have this problem. In a small net (like in my stiqueto-bot), i just used a couple a 'strategically' placed op-amps (my strategy involved poking it with a meter =). But in a net with a fan-out of 1000. This wouldn't be practical.
i read an interesting article in scientific american once. it mentioned how these neural net circuits would suddenly "remember" things as they were shut off. they talked about how it was like someone "seeing their life pass before their eyes" when they were dying. i wish i could remember the exact reference, it was very interesting. anyone else remember it?
i like german girls. and nannies.
This is one that is near and dear to my heart. As we move forward into this new digital age, we have to start looking at the issues that will begin to develop. Begin a huge Fan of Issac Asimove I have to think that his ideas and concepts of the future of Robots are probably some of the best around. He already saw some of the forth coming issues involved. In his works he created what are probably the best common sense laws of how it should all work that are possible. The laws of Robotics.
1) A robot may not injure a human being, or, through inaction, allow a human being to come to harm.
2) A robot must obey the orders given it by human beings except where such orders would conflict with the First Law.
3) A robot must protect its own existence so long as such protection does not conflict with the First or Second Laws.
Still as we move forward other issues will begin to come into play. As these contructs of man evolve so to speak, as the technology becomes greater will we not eventually have to reach a point that we must consider them more than a machine. Will we reach a point where they stop being tools and there by become a slave race?
That Idea gets explored again and agian in works of fiction. There are two stand out ST:THG episodes that explore just those ideas. I am not the sort of fan that remembers titles so you will have to forgive me.
In the first a scientist desires to strip data to his wires to find out more about him whereby be might make more Datas. Data fears this(in a way anyway, as he does not yet have emotions at this point) and is going to leave starfleet because he does not desire this, he fears he will lose the spark of life so to speak that makes him, himself. A ruling is made that Data is equipment, property to do with as Starfleet pleases("data is a toaster"). Picard and Riker are forced to argue opposite sides in a court case to descide what Data "Is". The thing that stands out the most for me in that episode is when Picard makes an impossioned speech and says to the effect "Our Misison is to seek out new life, well there it sits", He furthers the idea saying ok you repilcate Data make more of them, put them to work for Starfleet, use them for what ever purpose suits your fancy, sounds a lot like Slave labor. The descion is made that Data is infact a sentient being, with freewill, self awareness and even in a fashion the ability to reproduce.
The second happens a bit after that episode, the Enterprise is visiting a station where a scientist is using small Driods to repairs and maintain a project, they appear to be malfuntioning. Data thinks differently that they are acting in an intelligent manner, refusing to do dangerous or self destructive tasks. In try to prove this they setup a test that the driods see right through and work around seemingly failing yet in the end proving Data's theory. Data is able to prove that they are another form of intelligence much like himself.
Fianlly back to Asimov, he has two works that really prove out these Ideas, "I Robot" and "The Positronic Man", In 'I Robot' the Illistrated screenplay(A movie that seriously needs to get made but likely never will), we see how Robots develop, and move forward slowly becoming part of society, to the point where one becomes indistingushable from a human being. "The Positronic Man" takes this still further as the main Character the AI, wishes to become more than he is, he desires to grow beyond his original design, in the end becoming very human, to the point of creating an artifical body for himself. These works are incredible read them if you have not I can't begin to explain in this forum how they have shaped my thoughts on this subject.
Power Corrupts,Absolute Power Corrupts Absolutely, leaving one person(group)in charge is absolutely corrupt.
http://web.mit.edu/newsoffice
http://zig.ini.unizh.ch - The Zurich Institute of Neuroinformatics homepage.
This post brings up a question that I've been mulling over for a while. Perhaps those neurologists out there can answer this, because I'm completely at a loss.
Computer memory works with digital signals stored on a drive. You can look at that drive, and assuming that no access, etc. is going on at the time, those bits/bytes will stay in the same place, i.e. - they are static.
Now the human brain works almost nothing at all like a computer. Neurotransmitters cause K/Na gates to open, action potentials race down the neuron, which in turn releases its own neurotransmitters (if the threshold has been reached), and so on. So in the brain, there is NO static info-component! As far as I know, you can't just tell a neuron "Hold that action potential for me" (i.e. - a "1" on a drive).
That setup brings me to my question: How the heck does memory (short- and long-term) actually work ! For example, I can remember one particular biking accident on my my driveway when I was a kid. To some extent, I can even recall the smell of cut grass as I whipped around the corner and lost control.
Since there are no neurons "holding" that information, how has it remained in my brain for 12 years? Is there some sort of complex neuron-loop the information is going through in order to retain it? When I think about the memory, and "remember" it, how does the conscious portion of my brain access that memory and bring the incredibly detailed signal (still intact) to the surface?
This brings up another question about the storage limitations of the brain. There is an awful lot of information involved in the bike accident memory, and I have countless memories that are just as detailed. Is it possible that one stores concepts, rather than the fully-formed sensations (to save space)? For example, when I remember the smell of the grass from that long ago, the actual long-term signal is "smell of grass" and my brain accesses short-term memory to deliver that into the experience of remembering?
4-star general in a one-man army.
I'm sorry to repost this question, but I originally posted it under the wrong parent.
The original post brings up a question that I've been mulling over for a while. Perhaps those neurologists out there can answer this, because I'm completely at a loss. Computer memory works with digital signals stored on a drive. You can look at that drive, and assuming that no access, etc. is going on at the time, those bits/bytes will stay in the same place, i.e. - they are static. Now the human brain works almost nothing at all like a computer. Neurotransmitters cause K/Na gates to open, action potentials race down the neuron, which in turn releases its own neurotransmitters (if the threshold has been reached), and so on. So in the brain, there is NO static info-component! As far as I know, you can't just tell a neuron "Hold that action potential for me" (i.e. - a "1" on a drive). That setup brings me to my question: How the heck does memory (short- and long-term) actually work ! For example, I can remember one particular biking accident on my my driveway when I was a kid. To some extent, I can even recall the smell of cut grass as I whipped around the corner and lost control. Since there are no neurons "holding" that information, how has it remained in my brain for 12 years? Is there some sort of complex neuron-loop the information is going through in order to retain it? When I think about the memory, and "remember" it, how does the conscious portion of my brain access that memory and bring the incredibly detailed signal (still intact) to the surface? This brings up another question about the storage limitations of the brain. There is an awful lot of information involved in the bike accident memory, and I have countless memories that are just as detailed. Is it possible that one stores concepts, rather than the fully-formed sensations (to save space)? For example, when I remember the smell of the grass from that long ago, the actual long-term signal is "smell of grass" and my brain accesses short-term memory to deliver that into the experience of remembering?
4-star general in a one-man army.
The ability of humans to understand speech in noise is amazing. We're even able to understand speech when the speech *is* noise.
What I'm referring to is an experiment performed a few years ago, in which researchers recorded phrases and cut the recordings into small chunks. For example, 1/25 second, 1/15 second, 1/10 second, 1/4 second, 1/2 second etc. chunks. They then reversed each chunk, but left them in the original order. They then had people listen to those phrases with different chunk-lengths, and try to decipher what was being said. It turns out that if the chunk lengths are 1/15 second or less, you can still understand the message.
A few friends and I actually tried this out, and it's eerie. You know that those chunks have been reversed, but at 1/20 second, it sounds almost exactly like the original recording.
4-star general in a one-man army.
Geez, I'm just saturating you with questions, here. Sorry if this is getting to be too much, but I guess I'll ask anyway.... Is the mechanism utilized to change neuronal connections known?
Do neurons simply extend new processes and atrophy others to change existing shape/connections (ie - a multipolar neuron becoming a bipolar neuron or vice versa), or do entire neurons become apoptotic to make way for new connections?
4-star general in a one-man army.
I saw a talk by one of the authors, Rahul Sarpeshkar, who had just finished a postdoc at Bell Labs and was about to join MIT. Brilliant guy! I recall that in his talk he made an argument that the brain must use digital as well as analog processing. IN a nutshell, he argued that if the brain used purely analog processing, it would be defeated by noise. This came from an electronic cochlear and associated electronics that he built at Bell Labs. A number of analog amplification stages were used until the noise floor threatened to obliterate the signal. The signal was then digitised and the noise floor removed, then more analog stages etc. His argument used statistical noise, so it was nothing to do with electronics per se...
Yes, I'm only in second-year of an undergrad CS/Cognitive Science degree, but I seem to remember some of the reading I've done sort of contradict this.
Is it not true though that after the cell reaches its threshold and fires, if the activation potential of the neuron continues to increase, the neuron fires more often? Each firing is digital, but the frequency is analog.
It would seem to me therefore that every neuron is both a digital and analog device. Strange.
"Free beer tends to lead to free speech"
You sound like the emporer of China, right after the Great Wall had been finished, and when about to execute the fellow who had the audacity to invent a flying machine.
That said-- eralch said:Is it not true though that after the cell reaches its threshold and fires, if the activation potential of the neuron continues to increase, the neuron fires more often? Each firing is digital, but the frequency is analog.
Mostly analog. What I said was: Ability to reach threshold and probability of release can both be tuned by feedback loops which change the electrical properties of the cells, or which enhance or depress the chance of transmitter release. In different parts of the brain there are different mechanisms involved. By this I mean two things: First is is just what I said, that things like firing threshold and frequency can be tuned by feedback inputs, both from other neurons and from the neuron onto iteself. Second is that once you figure out the feedback mechanisms in one part of the brain it does not follow that those mechanisms are globally true. If the neuron in question doesn't have receptors for dopamine, it doesn't matter how much dopamine you dump on the cell.
And egerlach ended with: It would seem to me therefore that every neuron is both a digital and analog device. Strange.
Strange, indeed. In the brain it seems to be a combination of the two, but mostly analog, imo.
My prejudice is that because individual synapses are so damn complicated and variable, then the idea that anyone can claim to know how entire systems in the brain work makes me laugh behind my hand. On the other hand, what I study may seem chaotic and non-linear, but when viewed from more of a distance the sum of activity might resolve into apparent linearity.
The "tentacles", "onion", etc. Chakotay talks about are the dendrites, cell body and axon. These can be modeled quite effectively using cable properties, and the inputs he mentions can be described as variable resistors, etc. The math of just one multi-branched neuron can be reasonably complex, and putting together a model of a system requires some simplification. Computational neural networks are not constrained by the biology, which is often wet and squishy and multi-dimensional.
There are many ways in which the brain is both digital and analog. The misapprehension of the synapse as yes/no digital is based not on synapses in the brain, but on the connection between nerve and muscle. The neuromuscular junction has many fail-safes built in.
One aspect of neurons is certainly digital: Fire or not fire. There is a threshold based on the biophysics of voltage-gated channels. That same set of biophysical characters limit the waveform of the action potential (what we measure as a spike or "fire").
One aspect of neurotransmission is certainly stochastic: Transmitter release in certain areas of the cortex is not guaranteed in response to an action potential. Under normal circumstances some researchers have figured an average probability of only 1 in 3 that transmitter will be released in response to a spike. Even if transmitter is released, there is no guarantee it will be sufficient to bring the target cell to firing threshold.
Ability to reach threshold and probability of release can both be tuned by feedback loops which change the electrical properties of the cells, or which enhance or depress the chance of transmitter release. In different parts of the brain there are different mechanisms involved.
Just the engineering of the individual connections that make up the networks in the brain is very complicated. The article refers to the brain making "digital decisions" of car/not car, but even that is the result of complicated webs of signal input and processing.
If anyone tells you they understand how the brain works, they are either God or self-deluded.
Considering how the brain passes along information makes me wonder how they can state that the brain works digitally. While I am by far not an expert on neural processes, I would assume that the electrical signals travelling along the length of the nerve cells are definitely analog in nature. Considering each electrical pulse translates to a massive array of neurochemicals to bridge the synapsis, a short burst digital signal doesn't seem to be able to contain enough information. I think that it would be very interesting if someone with a bit more anatomical knowledge would enlighten us on what the electrical pulses "look" like.
So think I can turn in my brain and get something with a bit more memory.
"its great to be an american geek and somehow make a living at it" - michael stipe
A baby thinks very digitally, the block is eithor in sight or else it doesn't exist. Developement comes with the realization that the block can still exist if it is no tin sight. I recently read (can't remember where..USA Today?, Readers Digest?) that the brain is still actively creating paths well into adolescence. So it may be that we start thinking alond digital lines but learn to apply "fuzzy logic" along more analogue lines with a digital base. The use of analogue thinking is what allows the etchings of M.C. Escher to be so much fun.
In a time of universal lies, Telling the Truth is a revolutionary act - George Orwell
Ahh, but regardless if how powerful your computer is -- it's pretty much stuck on the desk or in the rack. I don't think we'll see a machine given both full mobility and sensory capabilities AND such an advanced 'brain' until the implications have been better studied/argued/legalized.
On a more optimistic note, there's no reason to _assume_ that a mechanized sentient would be any less cognizent of both it's rights and our own than any human sentient...it, as we, learns ethics by experience.
Of course, some humans don't learn as well as others...
If any intelligence, mechanical or biological, can grow to understand consequences and the need in a society to coexist, it would likely be quite happy living aside us 'ugly bags of mostly water', not with its foot to our collective throat.
Xentax
You shouldn't verb words.
I think you are assuming a few things, and using some incorrect technology. Rather, the article uses them incorrectly and you perpetuate it. When I read the article, I had the same thoughts. What if the car is fuzzy/in darkness/not clear/etc. This is the same iwth all pattern recognition. Even looking at your mother's face, you (or your brain, rather) matches this image to memory, and you can say with 99% certainty that that face is your mother. but what if the face changes ever so slightly? then you are less certain. It _could_ be her .. hmmm, 75% maybe? Now onto terminology. Digital just means that a signal is brken down into a binary medium. You do not have "digital" decisions. For a very simple example, I can look at the image of a car and say "It's a car", "It's not a car", or "I'm not sure". A musically talented friend of mine and I used to have huge analog vs. digital arguments. His view was that analog is superior becasue you can express the subtle differences along a true sliding scale rather than a "digitized" scale. But this arguement hodls no weight, as being digital is not the same as a binary off/on. For example, in music, you could have an analog tone that goes from 10,000 Hz to 15,000 Hz, sliding smoothly up the scale. But you can create a sliding scale just as smooth using digital if you break the scale down small enough so that the ear cannot tell where one tone begins and the other ends. The first step might be "on" at 10,000, then off. The next might be "on" at 10,000.001 Hz, etc. To sum, when you are talking a digital vs. analog arguement of any kind, there becomes ABSOLUTELY no difference between the two on a small enough scale. Vesuvius_DC
Now, 10% of autistic people have "Rainman" abilities - massive mathematical powers, etc., and apparentrly the current theory is that theses autistics are merely missing the final "step" in calculating things like humans do - the can't get that final estimate which allows us to get by in society easily.
I had no idea that autistics aren't humans!
Maybe that explains Ray Babbitt's fascination with Judge Wapner.
Gotta fly Qantas.
It's sad that you have no brain to fill.
By the power vested in me by the United States Constitution, I declare you a fucking moron.
I think when they say think like humans, they are refering to the ability to use logic and solve problems. Reasoning is what differs humans from machines. They are not trying to create an artificial human.
@%%%%= ###Connection failed, insufficient grey matter.
AS with most attempts at replicating the human brain, this development requires several elements. An analog frontend, a digital transactor, and a quantum computing device that would be able to react to the various stimulli produced by the digital componenets.
@%%%%= ###Connection failed, insufficient grey matter.
Actually, all Godel says is that you (may) have to go outside of things to be able to describe them accurately and completely (or: derive them from completely internal axioms). But since we (can) evolve as individuals, and are (hopefully) only a step in an evolutionary sequence, we (as a species) may generate a "next" species that can completely understand us. In this context, an AI device would only count if it were to act on the information it generated. And I would argue that if we were to comprehend such a conclusion from any source (external or internal) such conclusion would constitute an evolutionary step.
"Everything should be made as simple as possible--but not simpler" --Albert Einstein
..last post?
Mr. Last Post
/joeyo
2^5
I'd say that the main flaw in Penrose's argument stems from the fact that he seems to be seeking religion rather than searching for scientific (ie. testable) explanations.
I like well-written books that propose alternative theories, but they've got to have some sort of solid framework and internal consistency to be worth reading. The Emperor's New Mind was great as long as Penrose stuck to reviewing previous science, but appalling thereafter. I don't recall ever having read a popular science book containing so much handwaving, copouts, and defeatism. He's desperate to prove that scientific investigation is dead in the water when it comes to the mind, it seems to me. Put that together with some of the mystical mumbo jumbo that appeared liberally and it all starts to add up to a personal search for his God and The Reason He Must Exist.
Bleh, a very disappointing read.
"The question of whether machines can think is no more interesting than [] whether submarines can swim" - Dijkstra
The relation between firing and other neurons noticing and caring is not digital. However the neuron has binary way of attempting to communicate - namely firing.
Cheers,
Ben
My usual seat in the cluetrain is at A HREF="http://pub4.ezboard.com/biwethey.ht
FWIW my understanding of neurons is based on conversations with my wife who happens to have a PhD in biology and is pursuing her MD. (Note, the combined PhD/MD is considered a weaker combination than separate "real" degrees.)
Now I grant that neural networks may be different. And there may be differences between neurons.
But I definitely know that your basic neuron sends a stronger signal by firing more often, not by firing more strongly. Ditto for nerves and sensation. Stronger sensations are caused by more rapid firing, not more intense firing.
Regards,
Ben
My usual seat in the cluetrain is at A HREF="http://pub4.ezboard.com/biwethey.ht
A neuron's life comes down to deciding when to fire. Fire/not fire is a binary decision. There are not different types of firing. You do or you don't.
OTOH the neuron's decision to fire is influenced by all sorts of things from the chemical balance, what other neurons have fired recently, whether it tends to fire with them, etc.
So a neuron makes a digital decision on analog criteria...
Cheers,
Ben
My usual seat in the cluetrain is at A HREF="http://pub4.ezboard.com/biwethey.ht
Sadly, work on neural networks still sometimes relies a lot on buzz. Nature, as a journal, seems particularly susceptible to this kind of science: what they publish has to be short and pithy.
The researcher I believe you are looking for's name is Adrian Thompson. His web page is here. There is also an article on Discover's web site, if you go to their archives section and search for "FPGA" in the _body_ of the article. The article is called "Evolving a concious machine" and is by Gary Taubes. (Surprisingly it is the only article that contains the word FPGA in its body!)
I haven't looked at his work in a while, but I'm sure he has done some cool things with his evolving hardware since 1998. I always thought that the most interesting part was that he didn't limit the evolution to digital-only solutions-- resulting in incredibly efficient circuit designs that make *use* of crosstalk and interference!
Penrose's "The Emperor's New Mind" and "Shadows of the Mind" also makes the case (quite effectively, imo, but you may disagree) that the mind is not a digital computer, but a quantum computer and that to get a computer to think like we do we'll have to make it model quantum effects.
Admit nothing, deny everything and make counter-accusations.
The main flaw in Penrose's argument is that he gives no mechanism for the human brain to exploit quantum computation.
Of course, there are many other flaws, such as his assumption that humans aren't susceptible to "godelization". Sure we aren't susceptible to the SAME godel strings that number theory and turing machine are--that doesn't mean we are perfect.
Check out "Fabric of Reality" for a little more on this. There was another book I read recently that rebutted Penrose more effectively (and more thoroughly), unforunately I don't remember what it was.
--
Linux MAPI Server!
http://www.openone.com/software/MailOne/
(Exchange Migration HOWTO coming soon)
>I don't agree 100% that the brain makes digital decisions.
Not at the level of conscious thought, that is for sure.
However, at the level of the individual neurons, the response to a stimulus (whether to fire an action potential and at what frequency) are pretty much determined by the configuration of the cells and the electrochemistry of the cell membrane.
I think that the article just didn't make this clear.
LL
"If you are falling, dive." -Joseph Campbell
User:But...
Computer:Oh, and another thing, I was lying. I've seen much bigger hard drives.
Dreamweaver
"If a man hasn't discovered something he will die for, he isn't fit to live" -- MLK, Jr.
Somebody else made an excellent post describing this stuff in more detail.
To sum it up, the neuron acts pseudo-digitally. It must first determine if the stimulus is enough to fire. But once it has determined it is, it then fires an analog signal, the power of which is encoded by the firing rate.
Contrast this to a purely digital neuron, who, after recieving a stimulus, will just fire normally, and not really give any analog data as to how powerful the original stimulus was.
So real neurons are sort of passing some more info along, and I assume this allows for all sorts of subtle and nuanced feedback loops, etc., that may not be possible in a completely digital neural net.
It's 10 PM. Do you know if you're un-American?
Well, my impression is that if some level of stimulation hits an analog neuron, that neuron can fire others with some fraction of that stimulation. While a "digital" neuron would have to determine whether the signal was "enough" stimulation and if so, stimulate the others, otherwise don't stimulate them at all.
It's 10 PM. Do you know if you're un-American?
This was my first thought too. While the brain may make either-or decisions, that has no bearing on the actual nature of the process. Analog circuits can easily make "digital" decisions.
I think the problem lies in the author of the article. According to another post, the analog-digital thing happens on a neuron level. So the Yahoo article's explanation is just a bunch of hooey thrown in for those that won't question it.
How many times must we read about this kind of thing? We already know where it leads to: 1. Electronic circuit built to mimic human brain. 2. Circuit is put into super powerful Computer. 3. Computer reaches self-awareness and self-actualization. 4. Humans forced into servitude to the Computer. When will they ever learn?!?!- ----
--------------------------------------
Anyway, when he was done he had an FPGA that could tell the difference between "Stop" and "Go". The interesting part was that the program that it used wouldn't work on other FPGAs. Apparently, it was using analog effects that were specific to the individual chip. Furthermore, it was really efficient. Only a small percentage of the chip was being used. (Does this sound like your brain at all to anyone else?)
I was wondering if anybody had heard anything more about this research. I think it is facsinating.
First, we have a way to control smartness in animals, then we have a way to make electronics act like a brain. Combine those, and set the whole thing in flowers for algernon:
Janyuary 2023
My yuser think I stoppid, but I now I not be stoopid. How can I be stoopid wen I rite al these algo- alga- algarithims I think theyre caled. Yesterdy I rite a BSOD and my yuser no lik it. He say a nice peeple will help me and make me gen-yus. Just like Liynux. He is a mice with a jene to make him gen-yus. They saying they will do this too I. I hop I became gen-yus just lik he!
nuclear cia fbi spy password code encrypt president bomb
Friends don't let friends misuse the subjunctive.
Look, the important thing is not that we mimic the human mind or human thinking. Why do we want machines that think and act like humans? What good is that? So we can understand ourselves? Well, that is silly since the mechanisms that drive our intelligence are simply not going to be the same as the machines we make with human intelligence. That is, a computer with human intelligence tells us nothing about what really makes human intelligence actually work. The best a machine can do is ghost our cognitive economy, it cannot actually have it.
But that might be beside the point. More important is that we build and understand machines that have a higher level of intelligence than us. That intelligence might be nothing like a human's intelligence, but that's fine. As we all know, computers have a different kind of intelligence than us. And that is interesting. That should spark our creativity and that should get our juices going.
Here's an analogy. Suppose I build a telephone out of rubber bands and paper clips. It acts just like your favorite phone. But, is that interesting really? I mean, is the fact that we have a "really cool copy of a phone" all that interesting in terms of what-it-is-to-be-a-phone? Of course not. Instead, it is interesting that the damn thing is so complex and useful, even though it was made from rubber bands and paper clips.
Forget mocking the human experience. We get that each day, don't we? We get it (we're human). Let's look at other kinds of intelligences, based on machine mechanisms.
John S. Rhodes
WebWord.com -- Industrial Strength Usability
How to Download YouTube Videos
This invention is really a small step in the direction of having computers mimic the brain's capabilities at some cognitive abilities. For example, recently IBM showed that Big Blue, a computer, could beat the world's best at chess. The brain still has many areas at which it cannot be beat. Such as
Pattern recognition with translational invariance, rotation invariance, and size invariance
Speech recognition in noise.
Having computers that could perform things like pattern recognition or speech recognition as well as humans would allow enormous advances in the roles of humans and computers in our lives. People like Sebastian Seung think inventions like this will take them down that route - and ultimately result in huge scientific advances in artificial intelligence.
Personally, I think studying how the BRAIN does pattern recognition will allow far faster advances in this area than inventing chips that have SOME of the capabilities of neurons.
Neuromorphic electronic elements have been around a long time. There was an article from the 60's (in the Cold Spring Harbor symposia series, if I remember correctly) describing a vacuum-tube implementation of a neuron. I thought that was hilarious, given the hype that was going into the more recent silicon versions There is a basic design principle: don't do in hardware what you can do in software. Neuronal simulator packages like GENESIS and NEURON can simulate extremely realistic neuronal models in real time or faster. These neuronal models use 'compartments' which are little cylindrical segments of the cell, and they build up the geometry and electrical properties by putting them together. In other words, it is a spatial discretization of the cell. The finer the divisions, the more accurate the model. Last time I benchmarked, a PIII or Athlon class PC could handle something like 100 compartments in real time. That could either be 10 rather coarsely modeled neurons, or 1 very accurately modeled neuron. It is a great deal easier and cheaper to throw several PCs together in parallel to make a big network model, than it is to design an analog VLSI chip from scratch. The models are much more flexible, and can incorporate the latest data. You don't even want to think about what the debugging of VLSI would be like... I think Carver Mead's approach was the most practical: use the principles from real neural circuits, embody the equivalent computations in analog VLSI without being too picky about how neuron-like they were, and really use VLSI on a large scale. That is what he did with the artificial retina.
Anyone heard of the chaos theory here? Of course you have. Then you'll know that quantuum effects will propagate through all reality, not just even itself out over time. _Especially_ in the real world. (If you put things in a simulator, things _may_ converge on a grander scale, or not. It depends on the rules for feedback, what start-up conditions you begin with and what kind of number-system you use.)
So quantuum effects will always have _some_ effect. However, if it's big enough for dramatically changing how we think is another question. Alas, the whole sherade might not be so tied with our brain as we'd like to think either. Higher processing may manifest itself in quantuum effects in everything around us, including our whole body.
The problem is proving all this. Thank god everything can't be proved.
- Steeltoe
http://www.debunkingskeptics.com/
Regular neural networks still work on digital information only. These things, apparently, do not. That's why it's a big deal.
I do have a problem with this statement in the Wired article though:
Claiming that these guys have pioneered mixed-signal design is just a little bit of a stretch. Do your research, Wired. =)
--
Given a hand, a pencil, an eraser, and an infinitely long piece of paper, a brain can easily emulate a Turing machine. Does that make it a Universal Turing machine? (I think emulation is the only criterion). Even without the paper and pencil, it can emulate such a machine, apart from the poor storage capability.
Well, welcome to the real world! It's not always fair and sometimes people play unfairly to gain an advantage. What, do you think the media just snoops around MIT constantly looking for stories like this? MIT most likely has a good PR department (they'd be stupid not to).
Did the U of Manitoba do a press release on what they were doing? I didn't see it. Plus, this is apparently a joint corporate/university operation, so that's probably another reason we see it in the news.
Oh, and I don't mean to be cynical, but just because it happened doesn't mean you'll see it in the news -- another reality check for you.
As computer clocks go higher and higher, designers are going to have to become more and more aware that there's no such thing as a digital circuit. All electric and electronic circuits are analog.
When's the last time a computer designer had to worry about impedance matching between his circuit board and the components on it? As circuits become smaller and smaller electrically, transmission line effects become more and more important. Suddenly the digital designer finds that absolutely none of his signals are making it past the package leads due to lead inductance, or the dielectric constant of that cheap plastic package is high enough to cause the characteristic impedance of the line to be ten times lower than the PC board trace!
At least as an RF circuit engineer, my career is secure :)
Anyone who thinks that the human brain is a Turing Machine cannot consistently believe this sentence to be true.
-- the most controversial site on the Web
Biological (neural) systems have properties sometimes desirable electronically, such as robustness and insensitivity to noisy data. Indeed, Caltech's Carver Mead (if he's still there) went a long way to popularize biologically-inspired engineering, or "neuromorphic engineering." His book Analog VLSI and Neural Systems is the usual text, mixing VLSI design and mimicry of, say, the retina.
The original Nature article should be readable to those clued in on MOS circuitry and a bit of neuroscience. I think it's wonderful that Nature is willing to post their material for free online, esp. in PDF...
For those of you itching to learn more about the brain & neuromorphic engineering, I set up a page of links to related books.
All best,
Gregg Favalora, CTO, Actuality Systems, Inc.
Developing autostereoscopic volumetric 3-D displays.
Wired News offers a little more detail. The expressed difference is that it is a digital/analog hybrid. Apparently, the chip consists of standard transistors in a ring of artifical neurons and synapses. When impulses hit the neurons, they fire, but they can be regulated by a central inhibitor, blocking an ugly chain effect. The central inhibiting neuron allows control, including filtering of weaker signals to allow stronger ones to come through -- Sarpeshkar compares it to ignoring background noise at a party. It's an interesting concept, at any rate.
---
"Life. Don't talk to me about life."
The real challenge is going to be maintaining quantum states that are stable enough for work to get done. Right now standard computing is either on or off, and if something goes wrong, you just have to set up the computation again.
With quantum computing you may have a lot of work to reset, unless you can find a relatively easy way to generate quantum effects.
JHK
http://www.cascap.org and you'll never know unless you look
There was a conference at Stanford a while back (was mentioned here IIRC) on synthetic intelligence in general; all sorts of fun stuff was tossed out:
http://www.technetcast.co m/tnc_program.html?program_id=82
This quote (from John Holland) is particularly telling: So we're not quite there yet. Hans Moravec participated in the conference as well, and he has a fairly informative essay linked from his site entitled "When will computing hardware match the human brain?":
http://www.transhumanist.com/volum e 1/moravec.htm
:wq
Now, 10% of autistic people have "Rainman" abilities - massive mathematical powers, etc., and apparentrly the current theory is that theses autistics are merely missing the final "step" in calculating things like humans do - the can't get that final estimate which allows us to get by in society easily.
Are really cool machines that are trying to mimic humans ever going to get to stage where they can estimate things, or will they be like Data from Star Trek TNG. Hmmmm....
Acting stupid isn't much fun when there's someone around who knows better
I'm almost a little disappointed to read this coming from MIT, because when I left the University of Manitoba (Canada) a similar project was being given as a thesis project for fourth year students. The prof coordinating it has been doing research on building neural nets with semiconductors instead of software constructs for a while now. Granted, this bit from MIT might be more complex, or introduce new functionality to the neural net (such as the voice recognition system that incorporated time delays in the calculations last year). But it still seems to me that something is only big news if one of the 'big' colleges works on it. Bleah.
When I finish this internship and go back to finish my fourth year, I'll be proud to go to my hometown U. It's obviously keeping up with the rest of the world - the only thing lagging behind is the media's perception.
You know what to do with the HELLO.
You know what to do with the HELLO. ...
Help create an open-source world
http://www.wired.com/news/technology/0,1282,3702 9,00.html
I like how it's called a breakthrough in "neuromorphic" engineering. Doesn't it just become ten times more impressive when it's described in made up technomumbojumbo?
One time I threw a brick at a duck.
Yeah but could you make a...
Oh yeah, they're made to be clustered!
Eh...
I don't agree 100% that the brain makes digital decisions. The article says that we make an either/or decision regarding whether something is there or not. It is a car or it isn't a car. That's rather black and white. If a picture is blurry or if the object is partially hidden, then we could say, "It is almost a car," or, "It might be a car," implying that there is a degree to which something might be a car.
If you run an analog signal through a filter, you can detect if certain frequency is present. This may seem digital, similar to the car case, but actually it can be an analog signal and and analog filter. The results, similar to the car may be that the signal present, but it is not statistically significant above the background noise/interference.
To make a long story short; I still believe that humans make analog thoughts, even if our brain is just one big circuit.
Keeping
The Institute that is doing the research has more information here. I believe the guy doing the actual research has more research here.
Next time you get multiple submissions, try picking the post with more info than the rest instead of attempting to summarize. Especially when you leave out the important links.
--
Gonzo Granzeau
Gonzo Granzeau
"Nothing the god of biomechanics wouldn't let you into heaven for.." -Roy Batty
I've read the original paper in Nature. (I'd post a link, but I only have access via my university's account, and I have no interest in getting that revoked.) This is not exactly a neural network in the classic sense, although it is similar. The standard neural network is specifically designated an artificial network --- it implements a computational model of neurons. These guys are actually attempting to simulate the known electrical behavior of neurons, in the theory that a network composed of elements that truly mimic neurons will be more brain-like.
Now. "Digital and analog." This is not a new discovery. It has long been known that neurons have a specific threshold WRT to incoming signal; if the incoming signal does not meet the threshold, the neuron will not fire. If signal is above threshold, the neuron fires. If signal is really above threshold, the neuron fires repeatedly, encoding the strength of the stimulus as the frequency of the train of pulses. (AFAIK, the circuits described here didn't implement that last behavior.) This is a digital response. The output, however, is a continuous voltage at a particular frequency: an analog signal. (Whoever called this "a digital response to analog criteria" is correct.)
The important thing is that connections between neurons have different weights, and there's often a lot of local feedback. In practice, these feedback loops tend to be tuned so that a given cell will respond only to a fairly specific stimulus (the right light intensity in the right part of your visual field, or facing a certain direction relative to known landmarks, or hearing a sound from a certain direction, for example). These guys have implemented a circuit on silicon that shows the same filtering behavior and also captures the idea that neurons can be "on" or "off".
Yes, this is kind of neat. Yes, it could eventually lead to advances in AI; at the very least, it could provide useful signal filtering for robotic applications. No, it has nothing to do with plugging your Pentium into your parietal lobe or your Mac into your medulla, at least not until our circuit-design ability is so good that we can entirely mimic the black-box behavior of brain areas. (Hint: we don't even entirely understand that behavior for most regions.)
I'm also kind of surprised that this made Nature; there are guys at UPenn who've had working neuromorphic circuits for years now. Then again, it's only in the Letters section, and these new guys worked out some mathematical models for the gain of a neural circuit rather than just trying to copy existing ones.
sig not found
JHK
http://www.cascap.org